This study evaluated whether a deep learning-accelerated T2-weighted Dixon MRI scan could generate rapid, yet reliable, scans to diagnose degenerative lumbar spine conditions in patients with lower back pain. Traditional lumbar spine MRI protocols often require multiple scans, taking up to 7-9 minutes per scan, limiting scanner availability and increasing patient wait times. 30 adults with lower back pain underwent standard MRI protocols (sagittal T1-weighted and sagittal T2-weighted turbo spin-echo sequences) and an abbreviated single T2-weighted Dixon sequence employing deep learning reconstruction. Two radiologists (10 and 5 years musculoskeletal imaging experience) independently evaluated the two scans for common lower back conditions with a radiologist of 15 years experience serving as the diagnostic standard. Evaluation included determination of Modic changes, disc pathology, facet arthropathy, neuroforaminal stenosis, and Schmorl nodes. Compared to the standard scan, the abbreviated deep learning scan demonstrated an up to 84% reduction in scanning time, sensitivity of 100% for Modic changes and neuroforaminal stenosis, and a specificity of almost 100% for all pathologies. Between the readers, diagnostic confidence was comparable to the standard scans and inter-observer agreement was high, demonstrating the reliability of the newer scans between various experience levels.